diff --git a/src/process.h b/src/process.h
index 15403fd..33fd6bf 100644
--- a/src/process.h
+++ b/src/process.h
@@ -28,12 +28,12 @@ typedef struct st_process {
     uint64_t swap;
     uint64_t written;
     uint64_t read;
-
-    uint64_t childCpuTime;
-    uint64_t childMemory;
-    uint64_t childSwap;
-    uint64_t childRead;
-    uint64_t childWrite;
+    // cumulative fields - process + children
+    uint64_t cCpuTime;
+    uint64_t cMemory;
+    uint64_t cSwap;
+    uint64_t cRead;
+    uint64_t cWrite;
 } Process;
 
 #endif
diff --git a/src/server.c b/src/server.c
index 305a803..f2040df 100644
--- a/src/server.c
+++ b/src/server.c
@@ -59,6 +59,7 @@ int parseStatFile(Process *proc, char *filedata) {
 
   // (15) stime - %lu
   proc->cpuTime += fast_str2ull(&location);
+  proc->cCpuTime = proc->cpuTime;
   location += 1;
 
   // skip 16-19
@@ -81,12 +82,16 @@ int parseStatusFile(Process *proc, char *filedata) {
 
   proc->resident = 1024 * findAndParseField(&filedata, "VmRSS");
   proc->swap = 1024 * findAndParseField(&filedata, "VmSwap");
+  proc->cMemory = proc->resident;
+  proc->cSwap = proc->swap;
   return 0;
 }
 
 int parseIOFile(Process *proc, char *filedata) {
   proc->read = findAndParseField(&filedata, "read_bytes");
   proc->written = findAndParseField(&filedata, "write_bytes");
+  proc->cRead = proc->read;
+  proc->cWrite = proc->written;
   return 0;
 }
 
@@ -172,7 +177,60 @@ nextProcess:
 }
 
 void aggregateStats(Process *head) {
+  int processCount = resetVisits(head);
+  Process *current = head;
+  int depth = 0;
+  int maxdepth = 0;
+  int visited = 0;
 
+  // assign depth (aka level) of each process
+  while (visited < processCount) {
+    if (current == NULL) {
+      printf("Bad visit count: %d visited of %d processes\n", visited, processCount);
+      break;
+    }
+
+    current->visited = 1;
+    current->level = depth;
+    visited++;
+
+    if (depth > maxdepth) maxdepth = depth;
+
+nextProcess:
+    if (current->child != NULL && !current->child->visited) {
+      // process children first
+      depth++;
+      current = current->child; 
+    } else if (current->sibling != NULL) {
+      // process siblings next
+      current = current->sibling;
+    } else if (current->parent != NULL) {
+      // return to parent when tree is exhausted
+      depth--;
+      current = current->parent;
+      goto nextProcess;      // parent was already visited, so find next process from parent
+    } else {
+      // no parent - scan for unvisited process
+      while (current != NULL && current->visited == 1) current = current->next;
+    }
+  }
+
+  // scan for each depth level, tally stats upwards
+  for (depth=maxdepth; depth>0; depth--) {
+    current = head;
+    while (current != NULL) {
+      while (current != NULL && current->level != depth) current = current->next;
+      if (current == NULL) continue;
+
+      Process *parent = current->parent;
+      parent->cCpuTime += current->cCpuTime;
+      parent->cMemory += current->cCpuTime;
+      parent->cSwap += current->cSwap;
+      parent->cRead += current->cRead;
+      parent->cWrite += current->cWrite;
+      current = current->next;
+    }
+  }
 }
 
 char *readProcesses(char *procdir) {
@@ -256,10 +314,10 @@ char *readProcesses(char *procdir) {
   cur = cur->prev;
   free(cur->next);
   cur->next = NULL;
-
-  printFamilyTree(head);
+  
   linkFamily(head);
-  printFamilyTree(head);
+  aggregateStats(head);
+  // printFamilyTree(head);
 
   int clocks = sysconf(_SC_CLK_TCK);
   char *output = malloc(8 * 1024 * 1024);
@@ -272,14 +330,18 @@ char *readProcesses(char *procdir) {
             "pid=\"%d\",ppid=\"%d\",label=\"%s\",lxc=\"%d\"",
             cur->pid, cur->ppid, cur->label, cur->lxc);
     ptr += sprintf(ptr, "process_cpu_time_seconds{%s} %f\n", buffer, (double) cur->cpuTime / clocks);
-    ptr += sprintf(ptr, "process_child_cpu_time_seconds{%s} %f\n", buffer,(double) cur->childCpuTime / clocks);
-    ptr += sprintf(ptr, "process_num_threads{%s} %llu\n", buffer, cur->nThreads);
-    ptr += sprintf(ptr, "process_resident_bytes{%s} %llu\n", buffer, cur->resident);
-    ptr += sprintf(ptr, "process_swap_bytes{%s} %llu\n", buffer, cur->swap);
-    ptr += sprintf(ptr, "process_fileio_bytes_written{%s} %llu\n", buffer, cur->written);
-    ptr += sprintf(ptr, "process_fileio_bytes_read{%s} %llu\n", buffer, cur->read);
+    ptr += sprintf(ptr, "process_num_threads{%s} %lu\n", buffer, cur->nThreads);
+    ptr += sprintf(ptr, "process_resident_bytes{%s} %lu\n", buffer, cur->resident);
+    ptr += sprintf(ptr, "process_swap_bytes{%s} %lu\n", buffer, cur->swap);
+    ptr += sprintf(ptr, "process_fileio_bytes_written{%s} %lu\n", buffer, cur->written);
+    ptr += sprintf(ptr, "process_fileio_bytes_read{%s} %lu\n", buffer, cur->read);
     ptr += sprintf(ptr, "process_is_kernel_process{%s} %d\n", buffer, cur->iskernel);
     ptr += sprintf(ptr, "process_tree_depth{%s} %d\n", buffer, cur->level);
+    ptr += sprintf(ptr, "process_cumulative_cpu_time_seconds{%s} %lu\n", buffer, cur->cCpuTime);
+    ptr += sprintf(ptr, "process_cumulative_resident_bytes{%s} %lu\n", buffer, cur->cMemory);
+    ptr += sprintf(ptr, "process_cumulative_swap_bytes{%s} %lu\n", buffer, cur->cSwap);
+    ptr += sprintf(ptr, "process_cumulative_bytes_read{%s} %lu\n", buffer, cur->cRead);
+    ptr += sprintf(ptr, "process_cumulative_bytes_written{%s} %lu\n", buffer, cur->cWrite);
     // free and proceed
     Process *prev = cur;
     cur = cur->next;
@@ -361,7 +423,7 @@ int main(int argc, char *argv[]) {
   if (argc == 1) {
     char *buf = readProcesses("/proc");
     if (buf == NULL) return 4;
-    //printf(buf);
+    printf(buf);
     free(buf);
     return 0;